Nvidia’s $3,000 minicomputer steals the show at CES

Nvidia CEO Jensen Huang was received like a rock star this week at CES in Las Vegas, following an artificial intelligence boom that made the chip maker the second most valuable company in the world.

In his nearly two-hour speech Monday kicking off the annual conference, Huang filled a 12,000-seat arena, drawing comparisons to how Steve Jobs would unveil products in Apple the events.

Huang ended up with an Apple-like trick: a surprise product reveal. He introduced one of Nvidia’s server racks and, using some stage magic, held up a much smaller version that looked like a small computer cube.

“This is an AI supercomputer,” said Huang, wearing an alligator skin leather jacket. “It runs the entire Nvidia AI suite. All of Nvidia’s software runs on that.”

Huang said the computer is called Project Digits and works with a relative of Grace Blackwell’s graphics processing units (GPUs) that are currently powering the most advanced AI server clusters. The GPU is paired with one ARMGrace Central Processing Unit (CPU). Nvidia worked with Chinese semiconductor company MediaTek to create the system-in-one chip called GB10.

Formerly known as the Consumer Electronics Show, CES is usually the place to launch flashy and futuristic consumer gadgets. At this year’s show, which began on Tuesday and ends on Friday, several companies announced the integration of AI with appliances, laptops and even shutters. Other major announcements included a laptop from Lenovo that has a rotatable screen that can be expanded vertically. There were also new robots, including a Roomba competitor with a robotic arm.

Unlike Nvidia’s traditional gaming GPUs, Project Digits isn’t aimed at consumers. instead, it’s aimed at machine learning researchers, smaller companies and universities that want to develop advanced AI but don’t have billions of dollars to build massive data centers or buy enough cloud credit .

“There’s a gaping hole for data scientists and ML researchers who are actively working, who are actively building something,” Huang said. “Maybe you don’t need a giant cluster. You’re just developing early versions of the model and iterating over and over. You can do it in the cloud, but it just costs a lot more money.”

The supercomputer will cost about $3,000 when it becomes available in May, Nvidia said, and will be available from the company itself as well as some of its manufacturing partners. Huang said Project Digits is a placeholder name, indicating that it could change by the time the computer goes on sale.

“If you have a good name for it, contact us,” Huang said.

It’s a completely different kind of product than the GPUs that have fueled Nvidia’s historic boom in the past two years. OpenAI, which launched ChatGPT in late 2022, and other AI model makers like Anthropic have joined major cloud providers in buying Nvidia’s data center GPUs because of their ability to power the most intensive models and computing workloads.

Data center sales accounted for 88% of Nvidia’s $35 billion in revenue last quarter.

Wall Street has focused on Nvidia’s ability to diversify its business so that it is less dependent on a handful of customers buying massive AI systems.

“It was a little scary to see Nvidia come out with something so good for so little,” Melius Research analyst Ben Reitzes wrote in a note this week. He said Nvidia may have “stolen the show” because of Project Digits, as well as other announcements including gaming graphics cards, new robotics chips and a deal with Toyota.

Project Digits, which runs Linux and the same Nvidia software used in the company’s GPU server arrays, represents a major increase in capabilities for researchers and universities, said David Bader, director of the Institute for Data Science at the Institute of Technology. in New Jersey.

Bader, who has worked on research projects with Nvidia in the past, said the computer appears to be able to handle enough data and information to train the largest and most advanced models. He told CNBC Anthropic, Google, Amazon and others “would pay $100 million to build a training supercomputer” to get a system with these kinds of capabilities.

For $3,000, users can soon get a product that they can plug into a standard electrical outlet in their home or office, Bader said. It’s especially exciting for academics, who have often turned to private industry in order to get access to bigger, more powerful computers, he said.

“Any student that’s able to have one of these systems that cost about the same as a high-end laptop or gaming laptop, they’re going to be able to do the same research and build the same models,” Bader said.

Reitzes said the computer could be Nvidia’s first move into the $50 billion market for PC and laptop chips.

“It’s not too hard to imagine that it would be easy to do it all yourself and let the system run Windows one day,” Reitzes wrote. “But I think they don’t want to step on too many toes.”

Huang did not rule out that possibility when asked about it by Wall Street analysts on Tuesday.

He said MediaTek may be able to sell the GB10 chip to other PC makers in the market. He was careful to leave a mystery in the air.

“Of course, we have plans,” Huang said.

WATCH: Nvidia’s withdrawal due to CES expectations